Israelis develop 'self-healing' cars powered by machine learning and AI

Tel Aviv-based start-up Aurora Labs has made a proactive and remote system to detect and fix potential vehicle malfunctions, and update and validate in-car software without any downtime.

An illustration of Aurora Labs' technology (photo credit: AURORA LABS)
An illustration of Aurora Labs' technology
(photo credit: AURORA LABS)
Even before autonomous vehicles become a regular sight on our streets, modern cars are quickly resembling sophisticated computers on wheels.
Increasingly connected vehicles come with as many as 150 million lines of code, far exceeding the 145,000 lines of code required to land Apollo 11 on the Moon in 1969. Self-driving cars could require up to one billion lines of code.
For manufacturers, passengers and repair shops alike, vehicles running on software rather than just machines represent an unprecedented world of highly complex mobility. Checking the engine, tires and brakes to find a fault will certainly no longer suffice.
Seeking to build trust in the new generation of automotive innovation, Tel Aviv-based start-up Aurora Labs has developed software for what it calls the “self-healing car” – a proactive and remote system to detect and fix potential vehicle malfunctions, and update and validate in-car software without any downtime.
(From left) Aurora Labs co-founder & CEO Zohar Fox; co-founder & COO Ori Lederman; and EVP Marketing Roger Ordman (Credit: Aurora Labs)
(From left) Aurora Labs co-founder & CEO Zohar Fox; co-founder & COO Ori Lederman; and EVP Marketing Roger Ordman (Credit: Aurora Labs)
“The automotive industry is facing its biggest revolution to date,” Aurora Labs co-founder and chief operating officer Ori Lederman told The Jerusalem Post. “The most critical aspect of all that sophistication and software coming into the car is whether you can trust it, even before you hand over complete autonomy to the car. It poses a lot of challenges to car-makers.”
New challenges, Lederman added, include whether software problems can be detected after selling the vehicle, whether problems can be solved safely and securely, and whether defects can be solved without interrupting car use. In 2018, some eight million vehicles were recalled in the United States due to software-based defects alone.
“The human body can detect when something is not quite right before you pass out,” said executive vice president of marketing Roger Ordman. “The auto-immune system indicates something is wrong and what can be done to fix it: raise your temperature or white blood count. Sometimes the body can do a self-fix, and sometimes that’s not enough and needs an external intervention.
“Our technology has the same kind of approach – detecting if something has started to go wrong before it causes a catastrophic failure, indicating exactly where that problem is, doing something to fix it, and keeping it running smoothly.”
Advertisement
The company’s Line-Of-Code Behavior technology, powered by machine learning and artificial intelligence, creates a deep understanding of what software is installed on over 100 vehicle Engine Control Units (ECU), and the relationship between them. In addition to detecting software faults, the technology can enable remote, over-the-air software updates without any downtime.
Similar to silent updates automatically implemented by smartphone applications, Ordman added, car manufacturers will be able to update and continuously improve software running on connected vehicles. Of course, manufacturers will be required to meet stringent regulations, developed by bodies including the UNECE, concerning cybersecurity and “over-the-air” updates.
“When we joined forces and started developing the idea, we knew our technology was applicable to any connected, smart device or Internet of Things device,” said Lederman. “The first vertical we wanted to start with is the one that needs us the most, and the biggest market. The need for detecting, managing, recovering and being transparent about software is by far the largest need in the automotive industry as they move from mechanical parts to virtual systems run by lines of code.”
Rather than requiring mass recalls, Aurora Labs’ “self-healing” software will be able to apply short-term fixes to ensure continued functionality and predictability, and subsequently implement comprehensive upgrades to the vehicles’ systems.
The company, which has raised $11.5 million in fund-raising rounds since it was founded in 2016 by Lederman and CEO Zohar Fox, is currently working to implement its technology with some of the world’s leading automotive industry players, including major car-makers in Germany, the United States, Korea and Japan.
The fast-growing start-up also has offices in Michigan and the North Macedonian capital of Skopje, and owns a subsidiary near Munich.
“Customers ought to start being aware of how sophisticated their cars are,” said Lederman. “When they buy a new car, they should want to ask the dealership that they have the ability to detect, fix and recover so they don’t need to go the dealership. It’s something they would want to have.”
Just as the safety performance of cars in Europe are ranked according to the five-star NCAP standard, Ordman believes there should be an additional star for software safety and security.
“There should be as many self-healing systems in place as possible to enable that, when inevitably something does go wrong, there are systems in place to detect and fix them and maintain uptime,” said Ordman.
“Does the software running in the vehicle have the right cybersecurity in place? Does it have right recovery technologies in place? Can it continuously and safely improve over time?
“With these functionalities, you’re not just dealing with five stars of the physical but adding another star for the software safety and security. It is about giving the trust to the consumer. I’m getting a car that will safeguard me and my family as I move forward.”